REPORT ON THE PRESSURE ERRORS OF THE THERMOMETERS. 23 



In order to test the point I took one of the Challenger thermometers on which I had 

 already made numerous concordant experiments, and removed it from its vulcanite 

 sheath. After replacing the scale, I performed with it exactly the same experiments as 

 before. The result was unmistakable. The effect of pressure was notably diminished 

 by the absence of the vulcanite. This is a source of error which may differ greatly in 

 efficiency in the different thermometers, according to the qualitj 1 - of the vulcanite 

 and its exact position relatively to the protected bulb. 



The remaining part of the error, due to distortion of the glass protecting bulb, has 

 given me much more trouble than all the others together. Its amount cannot be 

 calculated, so far as I know, for it is a case of shear and compression combined, whereas 

 in the case of vulcanite it was a pure compression. I made some determinations, 

 however, by opening the protecting bulb, and substituting mercury and other liquids, 

 and sometimes air alone, for the alcohol which it originally contained. The general 

 conclusion from such experiments was, that a small amount of the whole observed effect 

 is due to the glass protecting bulb. 



To make this more certain I surrounded the protecting bulb with a test tube filled 

 with pounded glass, and I found the heating of this glass by compression, in spite of the 

 heating of the glass of the protecting bulb, produced a decided increase in the observed 

 rise of temperature. 



Thus it appears that there are no less than five different causes which contribute each 

 its share to Captain Davis' result. Of these, one is independent of the others, and would 

 produce its full effect even if they were not present. The other four give effects which 

 are not cumulative, and it would be very troublesome to try to assign to each its exact 

 share of the result when two or more act together. Fortunately, it will be seen that we 

 do not require to attempt to solve this problem. 



(1.) First is the direct effect of the external pressure upon the exposed part of the 

 thermometer tubes. This, in general, will be found very small, except in tubes where 

 there are large aneurisms. The whole effect of 3 tons pressure on a Challenger 

 thermometer without aneurisms, at temperatures near freezing point, so far as the 

 minimum index is concerned, would be only about 3 one thousandths of 30 degrees 

 or so, that is 90 thousandths or at most O'l of a degree for 3 tons pressure. That is an 

 amount which, in consequence of the necessary errors of reading the thermometers, may 

 be entirely neglected, and, unless there are large aneurisms, there will be little need 

 for pressure corrections even in six miles of sea. 



The other parts of the observed effect were 



(2.) Heating of water. This I observed to follow very nearly, according to Thomson's 

 formula, the original temperature of the water. By comparing the pressure effects on 

 the same thermometers during summer, and during winter (for which latter the late 

 continued frost was of particular service, and enabled me to work for many days at 



